ºÚÁϸ£ÀûÍø

Cooperative Missions of Autonomous Vehicle Swarms for Surveillance Tasks

Global project objectives: The ADVISOR project addresses a multifaceted problem in the autonomous vehicle (AV) industry – the lack of seamless interoperability between different classes of AVs: airborne, under- and on- water. The ADVISOR framework enables efficient development, testing, and execution of AV/swarm-based inspection systems. It provides capabilities to manage various aspects of intricate processes, workflows, and interactions, contributing to the early detection of issues, ensuring the reliable and safe operation of AV/swarms. With solutions that boost efficiency, cut costs, enhance security, and reduce environmental impact, the project promises substantial industry impact across sectors. The consortium comprises 21 partners from The Netherlands, Austria, South Korea, Turkey, Denmark and Portugal.

The Dutch consortium targets a teamed AV solution that is be able to autonomously inspect offshore wind turbines, pushing the SotA on AV autonomy, USV & UAV teaming and sensor data analysis/fusion. ºÚÁϸ£ÀûÍø (AIMS lab), Avular, Demcon, MU, DDC, Sorama and Vinotion join the forces to build a fully autonomous maritime system able to inspect offshore wind farms in North Sea.

Sensor partners Sorama and Vinotion focus on integrating advanced inspection functionalities, contributing by the development of smart acoustic sensors and AI vision technology for anomaly detection on wind turbine structures and blades specifically. This partnership enables autonomous and real-time monitoring of turbine structures, enhancing the detection capabilities for defects or anomalies.

DEMCON and Avular focus on raising the autonomy level of their water-borne and air-borne AVs, with research support from AIMS lab ºÚÁϸ£ÀûÍø on data processing, AI analysis, and sensor fusion. The integration of these technologies enables seamless collaboration between the drone and boat components of the teamed AV solution. Additionally, University of Maastricht contributes with a motion-compensated landing dock and accompanying drone and boat landing control and communication software, facilitating the safe and efficient landing of the Avular drone on DEMCON's USV under harsh offshore conditions. This will also enable the USV to perform as a carrier for the drone during the transit from shore to the wind park.

AIMS (AI for Multi-modal Sensing) lab targets the following innovations:

1. multi-modal SLAM algorithm for AV situational awareness, navigation and obstacle avoidance,
2. control system for robust autonomy of airborne and water vehicles in the rough maritime environments
3. AI-based windpark inspection and anomaly detection algorithms using combined video and acoustic sensor data.

Example Innovation: AI fusion of multi-modal sensor data for USV situational awareness offshore.

Currently, small and medium-sized USV boats are not able to efficiently handle offshore winds, waves, and currents. This is due to two main factors: small form factor (limiting the stability) and unavailability of inexpensive technologies providing all-round situational awareness and navigation in real-time. To tackle this challenge, we set up a sensor suite including standard GNSS, magnetometer, radar, LiDAR and high-res all-round cameras to develop an AI-based sensor fusion technology to obtain a more accurate and comprehensive understanding of the environment in real-time. We deploy the edge-near-sensor compute approach to partially process sensor data locally. This enables USVs to obtain timely information on current localization, orientation, obstacles around and make rapid decisions to adapt to changing conditions.

In situations where one of the sensor data inputs is compromised or temporarily unavailable, the multi-situational sensor data input system is designed to maintain its functionality by relying on the information provided by the other sensor systems. This redundancy and adaptability in the sensor architecture allow the system to make what can be described as wise decisions based on the available sensor data, ensuring continued safe autonomous navigation and complex collision avoidance path planning.

We target an IMO autonomy level 3 system, with features of level 4 involved.  While this already represents a significant advancement beyond the current state of the art, additional innovation lies in the realm of inter-drone communication, enabling precise flight path coordination for UAVs to safely land on USVs. Maintaining centimeter-level resolution for flight path determination when landing is considered essential in a GNSS-denied environment.